Site occupancy and abundance models for analyzing multiple-visit detection/nondetection data

Abstract

We propose an enhanced site occupancy model for analyzing ecological detection/nondetection data obtained from multiple visits. The model distinguishes between abundance, occupancy, and detection probabilities. We allow for transient individuals through a community parameter, c, that characterizes the proportion of individuals fixed across visits. This parameter seamlessly transitions from the standard occupancy model ($\mathit{c}=0$) to the N-mixture model ($\mathit{c}=1$), enabling a more accurate analysis of site occupancy data. Through theoretical developments and simulation studies, we demonstrate how this model effectively addresses biases inherent in conventional approaches, particularly for c is not at 0 or 1. We apply the model to various datasets of mammal and bird species and compare it to current approaches.